UCC2818-EP: UCC2818-EP PFC Input current problem

Hi Ramahadullah, 

Thank you for finding that paper and providing a link to it. Clearly your system is suffering from the slow-scale period-doubling instability that is investigated in the paper. (Note: Reference [6] in that paper refers to bifurcation period-doubling, but that refers to doubling of the switching period, not of the rectified AC line period. As I read it, your issue does not involve bifurcation.)
This paper tells me that my doubt about voltage-loop compensation involvement is incorrect. 

The input current of the PFC is shaped by the UCC2818-EP multiplier output signal, therefore one or more of the inputs to that multiplier must be carrying the slow-scale period-doubling modulation.   The three inputs to the multiplier are IAC, VFF, and VAOUT.  
IAC is a current, but is derived from the voltage on the output of the bridge rectifier.  As long as that voltage waveform is the same from half-cycle to half-cycle (as is normally expected), then IAC should also be the same each half-cycle of the line.
VFF is a voltage derived from IAC and should be nearly a DC level with a very small twice-line frequency ripple voltage superimposed on it. 

Please check and verify that, with the LLC stage attached, the rectified voltage and VFF voltages are as expected ( same as with resistive PFC load). 

Assuming that those are clean, that leaves the VAOUT waveform to carry the period-doubling modulation. 
Please check the VAOUT waveform to verify this.  

The paper discussed the interactions of the PFC control with that of a buck-derived DC-DC converter.  Your experience with an LLC second-stage indicates that the issue is more general (not specific to bucks) but we can suppose that the principles of stability should hold true regardless of DC-DC topology. 
On the other hand, the stability boundary curves of Figure 6 in the paper may be insufficient given that the LLC feedback loop gain-block may be much more complicated than a simple gain term over a time-constant term.  
Yet, the curves do suggest that stability can be achieved by adjusting one or more of the gain and/or crossover frequency parameters of either the PFC stage or the LLC stage, or both.

I think there is a possibility that this problem can be solved by incorporating some damping of low-frequency resonance between the two stages. 
Here are two papers by Erickson that I found in my files, that discuss adding damping to to improve the phase characteristics of the output and input impedances of two cascades stages.  The papers involve EMI filters interacting with the negative impedance of a buck DC-DC, but I think the concepts can apply to PFC with LLC as well.  
   Optimal_single_resistors_damping_of_input_filters.pdf      EMI Filter Damping.pdf      

Since damping an LLC converter may prove challenging, I suggest to first try adding R-C damping across the PFC output capacitor.  If that is successful, then try to achieve the same damping using R-L across the boost inductor.  But theoretically, the PFC is supposed to change the inductor effectively into a "current source" so L-R damping may not work.    

If any kind of damping does not work, then the fall-back plan is to change the gain/crossover characteristics of one or both of the power stages so that the period-doubling of the input current goes away. 

Regards,
Ulrich

Hi Ramahadullah, 

Did you try adding an R-C damping network across the PFC output cap?  Or increasing Cout by a large factor?  

After exhausting attempts to adjust the response of the PFC prove fruitless, I agree that the focus should shift to adjusting the LLC control. 

I am not an expert on LLC.  But I suggest that the same set of changes can be tried on the LLC control loop.  
-  Change loop gain
-  Change crossover frequency (either down or up). 
-  Try running the LLC from a benchtop DC source, if possible, to assess its stability.  

Regards,
Ulrich